Point-in-time feature join

Figure 01 / 12

The evaluated population

The cohort contains 18,200 decision snapshots, of which 491 have the defined synthetic outcome: Mature transaction loss. The outcome is known by construction here. In production, label uncertainty and selection must be recorded separately.

Figure data and text version

Outcome	Count
Mature transaction loss	491
Other labeled outcomes	17,709

A feature must contain only information available at the decision time. A late dispute cannot become an earlier transaction feature merely because its transaction date is old.

A training join uses the latest customer state and leaks future outcomes.

All amounts, rates, capacity limits, and outcomes in this case are synthetic. The three conditions are separate assumptions for comparison. A better result in the response condition is not measured proof that the proposed control causes that improvement. The figures expose the calculation and its limits; a real deployment needs its own evidence.

Read the result

The rule flags 1,210 of 18,200 decision snapshots. Of those flags, 236 meet the synthetic target, giving 19.5% precision. It misses 255 target events. Under the stated cost assumptions, residual loss and operating friction total $63,260. The important result is the connection between the population, action, capacity, and outcome—not one isolated score.

Figure 02 / 12

Four outcomes of the rule

The rule flags 236 synthetic positives and 974 negatives. It misses 255 positives. A flagged item is a decision to intervene; it is not proof of fraud, a legal prohibition, or any other real-world conclusion.

Figure data and text version

Known outcome	Flagged	Not flagged
Mature transaction loss	236	255
Other outcome	974	16,735

Figure 03 / 12

Three rates with different denominators

Precision is 19.5%, recall is 48.07%, and the false-positive rate is 5.5%. Changing the denominator changes the meaning. This record keeps each numerator attached to the population from which it came.

Figure data and text version

Metric	Numerator	Denominator	Result
Precision	236	1,210	19.5%
Recall	236	491	48.07%
False-positive rate	974	17,709	5.5%

Figure 04 / 12

Precision changes with prevalence

This sensitivity plot holds recall at 48% and false-positive rate at 5.5%, then changes prevalence. It is an algebraic comparison, not a forecast. Even unchanged detection quality can produce a very different review queue when the base rate changes. Horizontal positions are the labeled observations or scenarios; equal spacing does not imply equal numerical increments.

Figure data and text version

Assumed prevalence	Precision %
0.1%	0.87
0.5%	4.2
1%	8.1
2%	15.12
5%	31.48
10%	49.23

Figure 05 / 12

The threshold trade-off

Six illustrative score bands use a stated pair of detection rates. Lower sensitivity can reduce false alarms but miss more target events. These points do not come from a trained model and do not establish the best operating threshold. Horizontal positions are the labeled observations or scenarios; equal spacing does not imply equal numerical increments.

Figure data and text version

Score band	True positives	False positives
Band 1	481	2,656
Band 2	462	1,417
Band 3	422	620
Band 4	354	213
Band 5	246	71
Band 6	123	18

Figure 06 / 12

A transparent loss-and-friction calculation

At $210 severity per missed synthetic positive, residual loss is $53,550. Review costs $4,840; lost contribution on false alarms is $4,870. The calculation assumes intervention prevents all flagged-positive loss and each false alarm loses the stated contribution. Relax those assumptions before applying it to a real policy.

Figure data and text version

Cost component	USD
Missed-positive loss	53,550
Review cost	4,840
False-alarm contribution	4,870

Figure 07 / 12

Observed outcomes mature over time

The final synthetic positive count is 491. Earlier observations reveal only a stated fraction. Comparing a day-1 cohort with a day-30 cohort would confuse label age with control quality. This curve models observation delay only; it does not change the final outcome. Horizontal positions are the labeled observations or scenarios; equal spacing does not imply equal numerical increments.

Figure data and text version

Days after event	Observed positives
1	88
3	172
7	295
14	403
30	491

Figure 08 / 12

Review demand and available capacity

The flag count is 1,210. The comparison capacity is an illustrative 728 reviews per cohort window. A mathematical rule can be coherent while its resulting workload exceeds the operating team’s capacity. Capacity is not permission to ignore an applicable mandatory control.

Figure data and text version

Queue measure	Items
Flagged for review	1,210
Available capacity	728
Excess demand	482

Figure 09 / 12

A feature is an observation with provenance

This evidence contract supports make time-aware features reproducible. A value needs its event time, arrival time, scope, and source. Keeping unavailable evidence distinct from a measured zero prevents an outage from becoming a falsely reassuring feature.

Figure data and text version

Field	Example	Meaning
entity_ref	Point-in-time feature join	Subject of this case
event_time	2026-09-18T09:00:00Z	When the event occurred
received_time	2026-09-18T09:00:02Z	When the system learned it
signal_status	late	Evidence quality, not an outcome
label_definition	Mature transaction loss	The target used in these calculations

Figure 10 / 12

Missing evidence changes the observed population

The cells show an explicitly constructed completeness profile for three signal groups. The stress condition removes more history and device evidence. Missingness does not prove the target outcome; it changes what the decision process knows.

Figure data and text version

Signal group	Available	Missing
Identity evidence	16,016	2,184
Activity history	13,104	5,096
Context signal	10,920	7,280

Figure 11 / 12

Evidence, score, and action remain separate

The policy can use make time-aware features reproducible only within its approved scope. The action record must retain which evidence was available, which model or rule ran, and which action was actually applied. The final action can differ from the score recommendation when a separate constraint applies.

Figure data and text version

Stage	Record
Observe	Point-in-time feature join: evidence as of the decision time
Evaluate	Rule flags 1,210 of 18,200 decision snapshots
Apply	Record action, reason, owner, and expiry
Reconcile	Join the action to later outcomes without overwriting history

Figure 12 / 12

What the result cannot establish

Observed classifications do not reveal every counterfactual. The synthetic labels make arithmetic possible, but production decline data is selected by prior policy. Keep measured outcomes, assumed prevention, and unknown alternatives separate when reporting impact.

Figure data and text version

Claim	Evidence in this case	Limit
Detected target	236 known synthetic positives flagged	Production labels may be delayed or wrong
Prevented loss	Assumed 49,560 USD	Requires an intervention-effect assumption
Customer impact	974 synthetic negatives flagged	Not every flag causes abandonment
Unobserved alternative	Outcome without the action	Needs a valid evaluation design

Connect the result to the system

Join by both event time and availability time, then preserve the feature snapshot.

Check the population, evidence, permitted action, and actual effect together. A balanced calculation can still use the wrong population; a successful response can still leave an unknown financial outcome. The case’s numerical result applies only to its stated assumptions.